Datasets

With a little Google-fu, I managed to turn up data files for the Second Edition of Box, Hunter, and Hunter. I thought I might be able to scavenge the sets that were the same, but it turns out the second edition got a pretty thorough revamp, and most of the data is different. Darn.

With that, I added the data for all chapter two. Chapter Two was actually pretty interesting. George Box was actually a graduate student of Sir Ronald Fisher, and so he was close to the source. Box was very forthright about the limitations of statistics, with the statement, "All models are wrong. Some are useful." attributed to him. Chapter Two featured an industrial stats example that is completely unworkable if you use the standard statistical tests that require independence of observations. I liked this example, because it matches what often happens in industry, you simply don't have data that meets the standard assumptions of normality, independence, and random sampling, because your process just doesn't work that way. Of course, the stats courses I took at NAU were introductory, so they only taught techniques that depend on those three assumptions. I like knowing what to do when that won't work.